基于结构噪声信息的大型风力机传动系统运行状态识别与故障诊断方法研究

Large wind turbine drivetrain is difficult to be condition monitoring effectively, and operating efficiency is low. Aiming at this problem, research on condition monitoring and fault diagnosis method of large wind turbine drivetrain is proposed by this project based on structure noise information..First, feature extraction method for structure noise information of large wind turbine drivetrain is studied based on the blind separation and acoustic holography method. Second, condition monitoring method for spatial scale of large wind turbine drivetrain is studied based on the big data analytics theory. Third, fault diagnosis method of large wind turbine drivetrain is researched by the modern noise signal processing methods. Last, combining the phase space reconstruction and support vector machine theory, fault and operation safety prediction method of large wind turbine drivetrain are studied..This project is expected to elucidate feature extraction method for structure noise information of large wind turbine drivetrain; to reveal change law and mapping relation between condition/fault feature and structure noise of large wind turbine drivetrain, to achieve condition monitoring and fault diagnosis based on structure noise; to master improvement method of power generation capacity and operation safety prediction method of large wind turbine drivetrain. Research on this project is of important theoretical significance and practical value to change law of condition, fault and operation safety prediction by functional degradation in large wind turbine drivetrain.

针对大型风力机传动系统运行状态难以有效监控、运行效率低的问题，提出研究基于结构噪声信息的大型风力机传动部件运行状态识别与故障诊断方法。.项目将基于盲分离和声全息方法，研究大型风力机传动系统结构噪声信息特征提取方法；基于大数据分析理论，研究空间尺度大型风力机传动系统运行状态识别技术；基于现代噪声信号处理方法，研究大型风力机传动系统故障诊断方法；结合相空间重构和支持向量机，研究大型风力机传动系统故障预示方法及运行安全性预测方法。.项目预期将阐明大型风力机传动系统结构噪声信息特征提取方法；揭示结构噪声随大型风力机传动系统运行状态和故障特征的变化规律及其映射关系，实现基于结构噪声的运行状态监控与故障诊断；掌握大型风力机状态对于发电能力提高及运行安全性预测方法。项目的实施对于掌握风力机传动系统功能劣化引起的运行状态变化规律、故障预示与安全性预测具有重要的科学意义与应用价值。

以服役期内的兆瓦级风力发电机组为研究对象，建立了大型风力机传动系统关键部件在不同故障程度下的动力学模型，得出了系统内关键部件在极端载荷作用下的动力学响应。推导了子声源叠加的结构噪声识别方法，结合理论与实验，确立了风力机故障与结构噪声之间的映射关系。考虑交变载荷作用下风力机传动系统的运行状态，结合亥姆霍兹方程推导了子声源叠加的结构噪声识别方法，确立了风力机故障与结构噪声之间的映射关系，提出了利用结构噪声信息对大型风力机传动系统的运行状态识别方法。提出了改进的故障信号特征提取方法，建立了在非平稳、强噪声和非线性等复杂条件下的大型风力机传动系统的故障诊断策略，实现了在不同工况下对系统的关键零部件的故障特征进行准确地提取。实施了利用智能诊断方案对传动系统关键部件的不同类型故障的准确分类，完善了大型风力机传动系统运行状态识别的理论体系。以人工智能理论体系为基础，得到了系统内部件状态退化渐进演变的一般规律，给出了状态划分和趋势预测的普适性准则。考虑传动系统在运行过程中不同状态信息的劣化过程，构建了采用多源信息融合的传动系统剩余寿命预测理论，实现了对关键零部件的剩余寿命进行准确地预测。完善了相应的监测评价标准和运行状态预测标准，参考实验结果进行反复的修正，为风力发电机传动系统的状态劣化研究提供了一定的理论基础。通过兆瓦级风力机传动系统实验台与风电机组现场实验，充分考虑了传感器的类型与布置、转速的变化以及传动系统的故障类型，综合采用所推导出的诊断策略对实测信号进行分析，验证了提出的风力机传动系统故障预示及运行安全性预测方法。最终，利用所研究出的成果初步建立了大型风力机传动系统服役状态的评价准则。